• Rural Digital Europe
• Publications close
• LB close
• arXiv.org e-Print Archive close

Advances in deep learning and transfer learning have paved the way for various automation classification tasks in agriculture, including plant diseases, pests, weeds, and plant species detection. However, agriculture automation still faces various challenges, such as the limited size of datasets and the absence of plant-domain-specific pretrained models. Domain specific pretrained models have shown state of art performance in various computer vision tasks including face recognition and medical imaging diagnosis. In this paper, we propose AgriNet dataset, a collection of 160k agricultural images from more than 19 geographical locations, several images captioning devices, and more than 423 classes of plant species and diseases. We also introduce AgriNet models, a set of pretrained models on five ImageNet architectures: VGG16, VGG19, Inception-v3, InceptionResNet-v2, and Xception. AgriNet-VGG19 achieved the highest classification accuracy of 94 % and the highest F1-score of 92%. Additionally, all proposed models were found to accurately classify the 423 classes of plant species, diseases, pests, and weeds with a minimum accuracy of 87% for the Inception-v3 model.Finally, experiments to evaluate of superiority of AgriNet models compared to ImageNet models were conducted on two external datasets: pest and plant diseases dataset from Bangladesh and a plant diseases dataset from Kashmir. Comment: Accepted by Frontiers in Plant Science

The Internet-of-Things (IoT) is a revolutionary technology that is rapidly changing the world. IoT systems strive to provide automated solutions for almost every life aspect; traditional devices are becoming connected, ubiquitous, pervasive, wireless, context-aware, smart, controlled through mobile solutions, to name but a few. IoT devices can now be found in our apartments, places of work, cars, buildings, and in almost every aspect of life. In this investigation, we propose an IoT system Development Model (IDM). The proposed IDM enables the development of IoT systems from concept to prototyping. The model comprises concept refinement pyramids, decision trees, realistic constraint lists, architecture and organization diagrams, communication interface patterns, use cases, and menus of analysis metrics and evaluation indicators. The investigation confirms that the proposed model enjoys several properties, such as, clarity, conciseness, thoroughness, productivity, etc. The model is deployed for a variety of systems that belong to heterogeneous areas of application; the model is proven to be effective in application and successful in integrating mobile solutions. This chapter includes the presentation of the IDM submodels, the reasoning about their usefulness, and the technical developments of several systems. The chapter includes thorough discussions, analysis of the model usability and application, and in-depth evaluations. Comment: 31 pages, 16 figures, 3 tables

Extensive research in the field of monocular SLAM for the past fifteen years has yielded workable systems that found their way into various applications in robotics and augmented reality. Although filter-based monocular SLAM systems were common at some time, the more efficient keyframe-based solutions are becoming the de facto methodology for building a monocular SLAM system. The objective of this paper is threefold: first, the paper serves as a guideline for people seeking to design their own monocular SLAM according to specific environmental constraints. Second, it presents a survey that covers the various keyframe-based monocular SLAM systems in the literature, detailing the components of their implementation, and critically assessing the specific strategies made in each proposed solution. Third, the paper provides insight into the direction of future research in this field, to address the major limitations still facing monocular SLAM; namely, in the issues of illumination changes, initialization, highly dynamic motion, poorly textured scenes, repetitive textures, map maintenance, and failure recovery. A general guideline for the design of a Keyframe-based monocular SLAM.A survey on all Keyframe-based monocular SLAM (KSLAM) systemsInsight into the current state of KSLAM and future directions

Estimation, recognition, and near-future prediction of 3D trajectories based on their two dimensional projections available from one camera source is an exceptionally difficult problem due to uncertainty in the trajectories and environment, high dimensionality of the specific trajectory states, lack of enough labeled data and so on. In this article, we propose a solution to solve this problem based on a novel deep learning model dubbed Disjunctive Factored Four-Way Conditional Restricted Boltzmann Machine (DFFW-CRBM). Our method improves state-of-the-art deep learning techniques for high dimensional time-series modeling by introducing a novel tensor factorization capable of driving forth order Boltzmann machines to considerably lower energy levels, at no computational costs. DFFW-CRBMs are capable of accurately estimating, recognizing, and performing near-future prediction of three-dimensional trajectories from their 2D projections while requiring limited amount of labeled data. We evaluate our method on both simulated and real-world data, showing its effectiveness in predicting and classifying complex ball trajectories and human activities. Comment: Pattern Recognition, ISSN 0031-3203, Elsevier, 2017

Context. Gaia is an ESA cornerstone mission launched on 19 December 2013 aiming to obtain the most complete and precise 3D map of our Galaxy by observing more than one billion sources. This paper is part of a series of documents explaining the data processing and its results for Gaia Data Release 1, focussing on the G band photometry. Aims. This paper describes the calibration model of the Gaia photometric passband for Gaia Data Release 1. Methods. The overall principle of splitting the process into internal and external calibrations is outlined. In the internal calibration, a self-consistent photometric system is generated. Then, the external calibration provides the link to the absolute photometric flux scales. Results. The Gaia photometric calibration pipeline explained here was applied to the first data release with good results. Details are given of the various calibration elements including the mathematical formulation of the models used and of the extraction and preparation of the required input parameters (e.g. colour terms). The external calibration in this first release provides the absolute zero point and photometric transformations from the Gaia G passband to other common photometric systems. Conclusions. This paper describes the photometric calibration implemented for the first Gaia data release and the instrumental effects taken into account. For this first release no aperture losses, radiation damage, and other second-order effects have not yet been implemented in the calibration. 15 pages, 16 figures, Gaia data release 1 documentation special volume

The Gaia SpectroPhotometric Standard Stars (SPSS) survey started in 2006, it was awarded almost 450 observing nights, and accumulated almost 100,000 raw data frames, with both photometric and spectroscopic observations. Such large observational effort requires careful, homogeneous, and automated data reduction and quality control procedures. In this paper, we quantitatively evaluate instrumental effects that might have a significant (i.e.,$\geq$1%) impact on the Gaia SPSS flux calibration. The measurements involve six different instruments, monitored over the eight years of observations dedicated to the Gaia flux standards campaigns: DOLORES@TNG in La Palma, EFOSC2@NTT and ROSS@REM in La Silla, CAFOS@2.2m in Calar Alto, BFOSC@Cassini in Loiano, and LaRuca@1.5m in San Pedro Martir. We examine and quantitatively evaluate the following effects: CCD linearity and shutter times, calibration frames stability, lamp flexures, second order contamination, light polarization, and fringing. We present methods to correct for the relevant effects, which can be applied to a wide range of observational projects at similar instruments. Comment: 16 pages, 13 figures, accepted for publication in Astron. Nachr

J. Kr., A.M. and C.P. are grateful for support from DFG grant number KR 4015/1-1, the Carl-Zeiss Foundation and the Baden Württemberg Foundation. S. P., G. R., Q. F., C. F., K. P., S.C. and J.Ke. acknowledge support from the Presidential Innovation Fund of the Max Planck Society. G.R. was supported by an NSERC fellowship. J.G.S. acknowledges use of the Extreme Science and Engineering Discovery Environment (XSEDE), which is supported by NSF grant number OCI-1053575. E.B. and O.B. were supported by RFBR grants 13-06-00670, 13-04-01711, 13-04-90420. B.M. was supported by grants OTKA 73430 and 103983. A.Saj. was supported by a Finnish Professor Pool (Paulo Foundation) Grant. The Lithuanian sampling was supported by the LITGEN project (VP1-3.1-ŠMM-07-K-01-013), funded by the European Social Fund under the Global Grant Measure. A.S. was supported by Spanish grants SAF2011-26983 and EM 2012/045. O.U. was supported by Ukrainian SFFS grant F53.4/071. S.A.T. was supported by NIH Pioneer Award 8DP1ES022577-04 and NSF HOMINID award BCS-0827436. K.T. was supported by an Indian CSIR Network Project (GENESIS: BSC0121). L.S. was supported by an Indian CSIR Bhatnagar Fellowship. R.V., M.M., J.P. and E.M. were supported by the European Union Regional Development Fund through the Centre of Excellence in Genomics to the Estonian Biocentre and University of Tartu and by an Estonian Basic Research grant SF0270177As08. M.M. was additionally supported by Estonian Science Foundation grant number 8973. J.G.S. and M.S. were supported by NIH grant GM40282. P.H.S. and E.E.E. were supported by NIH grants HG004120 and HG002385. D.R. and N.P. were supported by NSF HOMINID award BCS-1032255 and NIH grant GM100233. This project has been funded in part with federal funds from the National Cancer Institute, National Institutes of Health, under contract HHSN26120080001E. This Research was supported in part by the Intramural Research Program of the NIH, National Cancer Institute, Center for Cancer Research We sequenced the genomes of a ∼7,000-year-old farmer from Germany and eight ∼8,000-year-old hunter-gatherers from Luxembourg and Sweden. We analysed these and other ancient genomes with 2,345 contemporary humans to show that most present-day Europeans derive from at least three highly differentiated populations: west European hunter-gatherers, who contributed ancestry to all Europeans but not to Near Easterners; ancient north Eurasians related to Upper Palaeolithic Siberians, who contributed to both Europeans and Near Easterners; and early European farmers, who were mainly of Near Eastern origin but also harboured west European hunter-gatherer related ancestry. We model these populations' deep relationships and show that early European farmers had ∼44% ancestry from a 'basal Eurasian' population that split before the diversification of other non-African lineages.